HAL Id: hal-02864936
https://hal.archives-ouvertes.fr/hal-02864936
Submitted on 12 Jun 2020
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Maria Kanakidou, J. H. Seinfeld, S.N. Pandis, Ian Barnes, Franciscus
Johannes Dentener, Maria Cristina Facchini, Rita van Dingenen, Barbara
Ervens, A. Nenes, C. J. Nielsen, et al.
To cite this version:
Maria Kanakidou, J. H. Seinfeld, S.N. Pandis, Ian Barnes, Franciscus Johannes Dentener, et al..
Or-ganic aerosol and global climate modelling: a review. Atmospheric Chemistry and Physics, European
Geosciences Union, 2005, 5 (4), pp.1053-1123. �10.5194/acp-5-1053-2005�. �hal-02864936�
www.atmos-chem-phys.org/acp/5/1053/ SRef-ID: 1680-7324/acp/2005-5-1053 European Geosciences Union
Chemistry
and Physics
Organic aerosol and global climate modelling: a review
M. Kanakidou1, J. H. Seinfeld2, S. N. Pandis3, I. Barnes4, F. J. Dentener5, M. C. Facchini6, R. Van Dingenen5,
B. Ervens7, A. Nenes8, C. J. Nielsen9, E. Swietlicki10, J. P. Putaud5, Y. Balkanski11, S. Fuzzi6, J. Horth5,
G. K. Moortgat12, R. Winterhalter12, C. E. L. Myhre9, K. Tsigaridis1, E. Vignati5, E. G. Stephanou1, and J. Wilson5
1Environmental Chemical Processes Laboratory, Dept. of Chemistry, University of Crete, 71409 Heraklion, Greece 2California Institute of Technology, 210-41, 1200 E. California Blvd., Pasadena, CA 91125, USA
3Dept. of Chemical Engineering, University of Patras, Patras, Greece
4Bergische University Wuppertal, Physical Chemistry FB C, Gauss Str. 20, 42119 Wuppertal, Germany 5Climate Change Unit, Institute for Environment and Sustainability, JRC, Ispra, Italy
6Istituto di Scienze dell’Atmosfera e del Clima – CNR, Italy
7Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort Collins, Colorado 80523, USA 8Schools of Earth and Atmospheric Sciences and Chemical and Biomolecular Engineering, Georgia Institute of Technology,
Atlanta, Georgia, 30332-0340 Atlanta, USA
9Dept. of Chemistry, University of Oslo, Oslo, Norway
10Div. of Nuclear Physics, Dept. of Physics, Lund University, Lund, Sweden 11LSCE, CNRS/CEA, Orme des Merisiers, 91198 Gif-sur-Yvette, France
12Max Planck Institute for Chemistry, Atmospheric Chemistry Division, Mainz, Germany
Received: 3 August 2004 – Published in Atmos. Chem. Phys. Discuss.: 28 September 2004 Revised: 1 March 2005 – Accepted: 12 March 2005 – Published: 30 March 2005
Abstract. The present paper reviews existing knowledge
with regard to Organic Aerosol (OA) of importance for global climate modelling and defines critical gaps needed to reduce the involved uncertainties. All pieces required for the representation of OA in a global climate model are sketched out with special attention to Secondary Organic Aerosol (SOA): The emission estimates of primary carbona-ceous particles and SOA precursor gases are summarized. The up-to-date understanding of the chemical formation and transformation of condensable organic material is outlined. Knowledge on the hygroscopicity of OA and measurements of optical properties of the organic aerosol constituents are summarized. The mechanisms of interactions of OA with clouds and dry and wet removal processes parameterisations in global models are outlined. This information is synthe-sized to provide a continuous analysis of the flow from the emitted material to the atmosphere up to the point of the cli-mate impact of the produced organic aerosol. The sources of uncertainties at each step of this process are highlighted as areas that require further studies.
Correspondence to: M. Kanakidou ([email protected])
1 Introduction
Atmospheric aerosols can scatter or absorb solar radiation, which modifies therefore the radiative balance of the atmo-sphere (IPCC, 2001). Black carbon heats the atmoatmo-sphere by absorption of solar radiation, whereas most organic aerosol components cool the Earth’s atmosphere. Fine aerosols have sizes close to wavelengths in the visible and are thus ex-pected to have a stronger climatic impact than larger parti-cles. In addition fine aerosols are transported far from their source regions and their climatic and environmental impact is, therefore, delocalized compared to the emission areas. Hydrophilic aerosols can act as cloud condensation nuclei (CCN), and thus have an indirect climatic effect through modification of cloud properties (Novakov and Penner, 1993; Novakov and Corrigan, 1996; IPCC, 2001). A number of studies have indicated that organic aerosol plays an important role in both the direct as well as the indirect aerosol forcing (Liousse et al., 1996; Cooke et al., 1999; Hansen et al., 1998; Penner et al., 1998; Lohmann et al., 2000; Jacobson, 2001; Chung and Seinfeld, 2002). However, there is an uncertainty of at least of a factor of 3 related to this forcing (IPCC, 2001) that deserves further study.
Organic material significantly contributes ∼20–50% to the total fine aerosol mass at continental mid-latitudes (Saxena and Hildemann, 1996; Putaud et al., 2004) and as high as
Fig. 1. Ratio of concentrations of the secondary organic aerosol
(SOA) to the total organic aerosol (OA) as computed by a 3-D global chemistry transport model for surface (upper panel) and zonal mean distribution (lower panel) for the month of July (Tsi-garidis, 2003; Tsigaridis and Kanakidou, 2003).
90% in tropical forested areas (Andreae and Crutzen, 1997; Talbot et al., 1988, 1990; Artaxo et al., 1988, 1990; Roberts et al., 2001). Significant amounts of carbonaceous aerosols are also observed in the middle troposphere (Huebert et al., 2004). A substantial fraction of the organic component of atmospheric particles consists of water-soluble, possibly multifunctional compounds (Saxena and Hildemann, 1996; Kavouras et al., 1998; Facchini et al., 1999a). Carbonaceous species that exist in the aerosol phase tend to be identified ac-cording to the manner in which they entered the particulate phase. Organic compounds that are emitted directly in partic-ulate form are referred to as Primary Organic Aerosol (POA). Note that the particulate matter in which these organic
com-pounds reside may contain non-organic comcom-pounds as well. Many gas-phase organic compounds undergo oxidation in the gas phase to yield products, generally oxygenated, that have sufficiently low vapor pressures that they will partition themselves between the gas and aerosol phases. Such com-pounds are often referred to as semi- or non- volatile, and when residing in the aerosol phase, as Secondary Organic Aerosol (SOA). Thus, in its common usage, SOA refers to that organic component of particulate matter that transfers to the aerosol phase from the gas phase as products of gas-phase oxidation of parent organic species. Other classes of aerosol organic compounds exist that do not fit neatly into these two categories. One class is organic compounds emitted into the atmosphere in vapor form, which subsequently condense into the aerosol phase without undergoing gas-phase chem-istry. Since these compounds can be identified with a par-ticular source, it seems most appropriate that they fall into the POA category. Another class of compounds are gas-phase organic species that are absorbed into cloud droplets and subsequently end up in the aerosol phase when the cloud droplets evaporate and leave residual aerosol. Again, the dis-tinction can be drawn as to whether the compound was emit-ted directly by a source or resulemit-ted from chemical process-ing in the atmosphere, in terms of its categorization as POA or SOA, respectively. Model studies (e.g. Pun et al., 2003; Kanakidou et al., 2000; Tsigaridis and Kanakidou, 2003) in-dicate that under certain circumstances the main fraction of organic aerosol can be of secondary origin, i.e. it is chemi-cally formed in the atmosphere (Fig. 1).
This applies also to the free troposphere where low tem-peratures favour condensation of semi-volatile compounds that have been chemically produced locally or elsewhere. This highlights the importance of secondary organic aerosol (SOA) for direct and indirect forcing. In addition, inclu-sion of SOA in climate models is needed since verification of aerosol calculations with remote sensed techniques (e.g. satellite/sun photometers) requires a full description of all aerosol components. In-situ measurements are often not able to discriminate between primary organic aerosol (POA) and SOA.
The processes that have to be considered in climate mod-els to account for the organic aerosol (OA) and its climatic impact comprise chemistry, physics and biology. Chemical processes include chemical formation and transformation of the OA by homogeneous reactions followed by condensa-tion or/and by heterogeneous reaccondensa-tions on particle surfaces or/and in clouds. Physical processes that determine OA mass and size distributions are emissions of primary OA and SOA precursors, followed by transport by advection, convection and diffusion, mixing with other particles by coagulation, evaporation and condensation of organic vapours as well as dry and wet removal (in cloud and below cloud scavenging). Emissions of primary organic particles and also SOA precur-sors can occur by various sources in the boundary layer and to a lesser extent in the free troposphere. The dry and wet
removal processes of OA depend on the water solubility and size of the particles containing these compounds. Finally, the simulation of the climatic impact of these aerosols will addi-tionally require the description i) of the water uptake by the particles, which depends on their hygroscopic properties, and ii) of the optical properties of the OA that also depend on the state of mixing of OA components with other aerosol com-ponents and are needed for the computation of the extinction of solar radiation.
Therefore, the following questions need to be addressed with regard to the OA and particularly of the SOA in the atmosphere:
– What are the SOA precursor gases? How important are
their emissions into the atmosphere? How important are the emissions of the primary OA?
– What are the main mechanisms of SOA formation? And
how much SOA is formed in the atmosphere?
– How important is organic nucleation on a global scale? – How can we simulate the partitioning of semivolatile
OA species between the gas and particulate phases? Can we assume thermodynamic equilibrium between the two phases?
– What are the hygroscopic properties of OA? How is OA
involved in the CCN formation?
– What are the optical properties of OA? And how are
they altered during ageing of the aerosols?
– How are aerosols mixed in the atmosphere? And how
does this mixing alter their chemical, physical and opti-cal properties?
– What are the responses of the climate system to changes
in organic aerosol?
During the last decade important advances were made in un-derstanding OA and its behaviour in the atmosphere. Dif-ferent studies have shown in the past years that biogenic hy-drocarbons play a significant role in the formation of tropo-spheric ozone and that even in urban areas with high anthro-pogenic emissions they still need to be considered in order to develop reliable strategies for the reduction of tropospheric ozone (Chameides et al., 1988; Roselle, 1994; Vogel et al., 1995; Atkinson and Arey, 1998). It has also been known for quite some time that the oxidation of monoterpenes in the troposphere plays a potentially important role in the gener-ation of secondary organic particulate matter (Went, 1960; Rasmussen, 1972; Trainer et al., 1987; Jacob and Wofsy, 1988; Andreae and Crutzen, 1997). Many biogenic hydro-carbons show much higher reactivity towards the important atmospheric oxidants OH, NO3 and ozone than the
anthro-pogenically emitted VOCs, which adds further to the sig-nificance of VOCs emitted from biogenic sources (BVOCs)
as a major potential contributor to global organic particulate mater (PM) formation (Atkinson, 2000).
In recent years BVOCs have been positively identified as precursor substances to the formation of SOA in the atmo-sphere (Kavouras et al., 1999; Pandis et al., 1992; Yu et al., 1999a, b). That BVOCs contribute to aerosol formation has now been amply demonstrated in different laboratory and outdoor chamber experiments (e.g., Palen et al., 1992; Zhang et al., 1992; Hoffmann et al., 1997; Griffin et al., 1999a; Barnes, 2004; Hoffmann, 2001; Jaoui and Kamens, 2003a and references therein). In spite of this there is still a paucity of, data on and understanding of, the composition and the properties of the aerosol formed from the gas phase photoox-idation of biogenic hydrocarbons (Christoffersen et al., 1998; Kavouras et al., 1998).
A number of review papers exist that are used as a start-ing point for the present review: In 1996, Saxena and Hilde-mann (1996) identified and estimated the solubilities of an extensive set of water soluble organic compounds that could be present in atmospheric particles. Atkinson et al. (1997) have reviewed VOC gas phase chemical mechanisms, in-cluding those leading to aerosol formation. Seinfeld and Pankow (2003) summarized kinetic knowledge on SOA for-mation and appropriate parameterisations. Kulmala (2003) described the procedure of formation and growth of parti-cles in the atmosphere. Jacobson et al. (2000) focused on an extensive presentation and discussion of OA measurement techniques.
The present paper aims to build upon these earlier reviews especially by including recently acquired knowledge in the area of Secondary Organic Aerosol. It also intends to exam-ine the OA problem from the point of view of climate mod-elling and define critical areas where additional knowledge is needed to reduce the involved uncertainties. In the next sections, the state-of-the-art of all the components needed for the representation of OA in a climate model is outlined: The emission estimates of primary carbonaceous particles and SOA precursor gases are summarized. The up-to-date understanding of the chemical formation and transformation of condensable organic material is outlined together with an overview of the SOA formation representation in global cli-mate models. Measurements of physical and optical prop-erties of the organic aerosol are summarized and needs for modelling studies are highlighted. The mechanisms of in-teractions of SOA with clouds are discussed. Dry and wet removal parameterisations in global models are outlined. Ef-fort is put into synthesizing this information to provide a con-tinuous flow from the emitted material to the climatic impact of the organic aerosol. The sources of uncertainties at each individual step of the overall process are highlighted as areas that require further studies.
2 Emissions of primary carbonaceous aerosols and of SOA precursors
This section evaluates the current knowledge and uncertain-ties of emissions of primary carbonaceous aerosol and of the volatile organic compounds (VOC) that can contribute to the chemical formation of organic particulate matter in the atmo-sphere.
2.1 Primary carbonaceous emissions: global and regional emission estimates
Sources of primary carbonaceous particles include fossil fuel burning (especially transportation and energy production), domestic burning (cooking and heating), and uncontained burning of vegetation (savannah and deforestation fires) and agricultural waste. There are a number of other types of primary carbonaceous material in the atmosphere such as viruses, bacteria, fungal spores and plant debris (Bauer et al., 2002) that may be relevant because they are effective ice nuclei (see compilation of laboratory data by Diehl and Wurzler, 2004). Their contribution to aerosol mass may be substantial (Wiedinmyer, 2004, and references therein), but due to their residence in the coarse aerosol fraction, their climate relevance is generally considered to be relatively low. Recently a significant marine POA source acting dur-ing the period of high biological activity has been identified in North Atlantic (O’Dowd et al., 2004). During phytoplank-ton blooms, progressing from spring through autumn, bub-ble bursting produces submicrometre particles enriched in insoluble and high molecular mass organic matter. Unfortu-nately there are no quantitative estimates of their global and regional emissions, and to our knowledge there are no stud-ies on their potential role as condensation sites for SOA. It is currently believed that fine organic particles offer surface and mass, on which SOA precursor gases may preferentially condense.
Note that black carbon (BC) is an operationally defined quantity and the use of BC should be accompanied by the method used for its measurement. In the present paper we will use the term BC since it is more relevant to climate change. Therefore, the compiled inventories are based on emission measurements with all types of sampling and anal-ysis methods introducing thus significant inconsistencies be-tween the emission inventories and the observations. At present most measurement techniques can not discriminate between the organic aerosol formed from biogenic and an-thropogenic precursor gases and the primary carbonaceous particles emitted from, mainly, pyrogenic processes.
A recent and extensive analysis of regional black carbon (BC) and organic particulate carbon (OC) emissions is pro-vided by Bond et al. (2004), using energy statistics for the year 1996. Global emissions of BC are estimated to be 1.6, 3.3 and 3.0 Tg C y−1 for biofuel, vegetation fires, and fos-sil fuel burning, respectively. For Primary Organic Aerosol
these numbers are 9.1, 34.6, and 3.2 Tg POA y−1. Main
un-certainties are connected to the choice of emission factors that depend on the fuel burnt and the type of combustion. Biofuel consumption for domestic use is the source category associated with the highest uncertainty due to the difficulty in getting reliable statistics. A recent study by Schaap et al. (2004) suggests that, at least in Europe, the BC emis-sions of this inventory may be underestimated by a factor of two. Novakov et al. (2003) estimated historical trends in fos-sil fuel BC emissions since 1875. These trends show rapid increase in the latter part of the 1800s, levelling off in the first part of 1900s and the re-acceleration in the past 50 years as China and India have been developing. These changes that have caused regionally large temporal modifications in aerosol absorption might be accompanied by similar trends in OA emissions. These possible emission changes and their climate impact need to be evaluated.
2.2 SOA precursor emissions
2.2.1 Mechanism and composition of natural SOA precur-sor emissions
VOCs are emitted into the atmosphere from natural sources in marine and terrestrial environments, as well as from an-thropogenic sources. A key study on global natural emis-sions was published by Guenther et al. (1995); hereafter called G1995, which is still the basis for later estimates of natural VOC emissions. On a global basis the emissions of biogenic volatile organic compounds (BVOCs), which are emitted mainly by vegetation, are estimated to exceed those from anthropogenic emissions (G1995: Guenther et al., 1999, 2000). Recently, Wiedinmeyer et al. (2004), hereafter W2004, provided an excellent review paper of emissions of organics from vegetation. Here we focus on those emissions and their uncertainties relevant for SOA formation.
Isoprene accounts for about half of all natural VOC emis-sions and is, on a mass basis, the dominant emitted biogenic VOC component. Estimated global emissions range between 250 and 750 Tg C y−1(W2004). However, isoprene is gen-erally not considered as a major producer of SOA. Very re-cent studies, however, detected the presence of humic like substances, glycol aldehyde and hydroxy aceton as well as methyltetrols indicating involvement of isoprene as source for SOA (Jang et al., 2003a; Claeys et al., 2004a, b; Limbeck et al., 2003; Matsunaga et al., 2003). Claeys et al. (2004a) proposed that a small (0.2%) fraction of all isoprene may be converted into SOA, corresponding to 2 Tg y−1 emissions.
This number deserves revision since fastly after Claeys et al. (2004b) suggested that aqueous phase oxidation of iso-prene products is a more important source of SOA. Thus far more than 5000 terpenes have been identified (Geron et al., 2000), such as monoterpenes (C10), sesquiterpenes
(C15), diterpenes (C20)and higher molecular weight
Table 1. Mass percentage of monoterpene, and reactive ORVOC emission as given by Seinfeld and Pankow (2003; SP2003 based on
Guenther et al., 1995, and Griffin et al., 1999b), Owen et al. (2001) and Geron et al. (2000).
Species Mass % Contribution
Class SP2003 Owen Geron
global S. Europe and N. America Mediterranean
α-pinene M 24.8 30–58 12–53
β-pinene M 16.4 8–33 10–31
Sabinene+terpenoid Ketones M/ORVOC 10.0 2.5–14 2–5
13-carene M 3.0 0 4–9
Limonene M 16.4 0–5 6–10
α-γ terpinene M 0.6 2–5 0–6
Terpinolene M 1.4 n.d. 0–2
Myrcene M 3.5 0–4 2–7
Terpenoid alcohols ORVOC 14.9 0–20 n.d.
Ocimene M 1.5 0–1 0–1
Sesquiterpenes ORVOC 7.4 n.d. n.d.
1M: Monoterpenes;2ORVOC: Other reactive VOCs;3n.d.: not determined
after isoprene that is a hemi-terpene, are the mono-terpenes (C10H16) α-pinene, β-pinene, sabinene, and limonene
(Ta-ble 1), accounting for 40–80% of the overall terpene emis-sion on a global scale when isoprene is excluded. Field mea-surements have shown that the mono-terpenes represent a significant fraction of the BVOCs emitted from vegetation to the atmosphere with contributions ranging from 10 to 50% dependent on the type of vegetation prevailing in the area (G1995; Guenther, 1995; Guenther et al., 2000).
Excluding isoprene and methane, VOCs from biogenic sources are often divided (G1995) into the lumped cate-gories i) terpenes, ii) other reactive VOC (ORVOC) and iii) other VOCs (OVOCs). In the widely used GEIA dataset (URL http://geiacenter.org/), the latter two are lumped to-gether. ORVOC represent reactive VOCs, with lifetimes <1 day, such as terpenoid alcohols, n-carbonyls, aromat-ics, sesquiterpenes (C15H24), terpenoid ketones and higher
olefins. OVOCs are the less reactive VOCs, with lifetimes longer than 1 day, typically methanol, various aldehydes and ketones. The latter are believed to have little aerosol for-mation potential, and are not further considered in this sec-tion. Note however that recent studies report that >C7 car-bonyls may be important contributors to SOA (Matsunaga et al., 2003). Current analytical methods may need to be im-proved before we can accurately quantify these compounds. According to Griffin et al. (1999b), only about 30% of the lumped ORVOC and OVOCs have the potential to form SOA. In contrast, the mono-terpenes, and especially the sesquiter-penes (100%), have large potential for SOA formation. Note also that lumping of various chemical compounds in the emissions inventories like for instance ORVOC puts together compounds that form aerosols with compounds that are not
precursors of SOA. This introduces further uncertainties in the SOA modelling.
Woods, crops and shrubs contribute by 55%, 15%, and 14%, respectively, to the non-isoprene biogenic emissions (G1995), whereas oceans emit <1%. Emission amounts and composition are species and thus ecosystem dependent; the main external factors influencing emissions are i) tempera-ture ii) light (for some species) and iii) water stress. G1995 describe an algorithm that uses ecosystem input data, emis-sion factors, light and temperature dependent functions and a canopy radiative transfer model. The most recent up-dates of the algorithms are provided by the MEGAN activity (http://cdp.ucar.edu).
Seinfeld and Pankow (2003) combined the Griffin et al. (1999b) ORVOC breakdown in smaller chemical cate-gories and classify the SOA forming compounds in a total of 11 categories, which encompass on a global scale most of the observed terpenes and ORVOCs emissions, with the Guen-ther et al. (1995) mass emissions. For comparison we also give the fractions obtained in N. America (Geron et al., 2000) and Southern Europe Mediterranean (Owen et al., 2001). As becomes apparent in Table 2, the species contributions to the emissions adopted in global models is rather similar to those obtained by regional estimates. However, care should be taken in extrapolating regional results to the global scale, since precursors like sesquiterpenes that have the largest po-tential in forming SOA (e.g. Vizuete et al., 2004) have also highly uncertain emissions.
2.2.2 Anthropogenic SOA precursor emissions
Aromatic components have also the potential to form SOA (Odum et al., 1997). Tsigaridis and Kanakidou (2003)
Table 2. Regional breakdown of Anthropogenic Primary Organic
Aerosol, Black Carbon, terpenes (excluding isoprene), other reac-tive VOC (ORVOC) terpenes [Tg y−1]. Table adopted from Bond (2004).
Region POA BC terpenes ORVOC
OPEN OCEAN 0.1 0.03 0.2 2.8 CANADA 1.0 0.1 4.8 4.1 USA 1.9 0.4 8.3 13.6 LATIN AMERICA 10.5 1.3 48.5 104.4 AFRICA 16.8 2.0 28.2 57.0 OECD EUROPE 1.3 0.4 2.3 3.9 E. EUROPE 0.4 0.1 0.5 1.2 CIS (FORMER) 2.0 0.3 6.6 7.7 MIDDLE EAST 0.5 0.2 0.9 1.7 INDIA REGION 3.7 0.8 6.0 16.4 CHINA REGION 4.7 1.7 6.5 14.1 EAST ASIA 2.2 0.5 8.7 25.5 OCEANIA 1.6 0.2 5.7 7.8 JAPAN 0.1 0.2 0.4 0.6 WORLD 46.9 8.0 127.4 260.7
adopted the EDGAR2.0 database (Olivier et al., 1996, 1999a) for the anthropogenic emissions of SOA precursor gases. This database is giving global emissions of 6.7 Tg y−1 toluene, 4.5 Tg y−1xylene, 0.8 Tg y−1trimethylbenzene and 3.8 Tg y−1 of other aromatics. These emissions add up to about 10–15% of all anthropogenic NMVOC emissions.
These values were determined for the year 1990; emission factors are highly uncertain, and moreover subject to strong temporal changes. e.g. in Europe and USA decreasing trends of hydrocarbon emissions have been reported in the last 2– 3 decades. In the USA, reported NMVOC emissions have been decreasing from 35 kT y−1in 1970 to 20 kT y−1in 2001
(http://www.epa.gov/ttn/chief/trends/index.html). From sec-tor analysis it follows that traffic, industrial processes and solvent use were responsible for that large decrease. In Eu-rope, reported emissions decreased in the EU from 16 kT y−1 in 1989 to 12 kT y−1in 2000 (EMEP, 2003). Latter trends are confirmed by a limited number of measurements (Monks et al., 2003). Reductions have been reported for acetylene, ethane, benzene and toluene (Roemer, 2001). In contrast, in the period 1970–2000, in South and East Asia and China an-thropogenic NMVOC emissions may have increased by 50 % (IIASA, M. Amann, personal communication, 2004) from 41 to 63 Tg y−1.
2.2.3 Global and regional estimates
A gridded compilation of the global emissions, divided into two categories, terpenes and the lumped ORVOC and OVOC, has been made by G1995. Table 2 gives the regional break-down of the primary anthropogenic organic aerosol (POA)
flux estimates by Bond (2004) and terpene and ORVOC emissions. Note that the importance of oceans (O’Dowd et al., 2004) and of vegetation as natural sources of POA re-mains to be determined. As a rough estimate, assuming that a constant fraction of 0.15 of the terpene emissions reacts on a very short timescale to form SOA then 19.1 Tg y−1of SOA are calculated to be globally produced by terpenes. This crude assumption provides an order-of-magnitude compar-ison of the relative importance of SOA versus primary or-ganic aerosol emissions on regional and global scales. This estimate shows that the SOA contribution to OA is likely to be highly variable ranging from 10–70%, in Eastern Europe, and Canada, respectively. Obviously this approach should be viewed with caution, since as discussed below, SOA forma-tion is a complex and not yet sufficiently understood process. In addition, Tsigaridis and Kanakidou (2003) have shown that a significant proportion of the SOA formation occurs in the free troposphere due to enhanced condensation favoured by low temperatures. This is crucial for SOA fate since the lifetime of aerosol is larger in the free troposphere than in the boundary layer as discussed in Sect. 4.3.
We can make a similar, but even more speculative, analysis of the importance of ORVOC as precursor for SOA. If 30% of the ORVOC emissions can form SOA (of which 5% con-sist of sesquiterpenes), assuming a 100% aerosol yield for the sesquiterpenes (higher than reported by Griffin et al., 1999b), and 15% for the other components, an additional amount of 15 Tg SOA y−1could be formed, with regional contributions varying between 10 and 50%. Following the hypothesis by Claeys et al. (2004a) that a small but significant fraction of isoprene oxidation products may lead to SOA, an additional amount of 2 Tg SOA y−1, may be formed, 65% of which can
be attributed to Africa and S. America. In addition, anthro-pogenic organic compounds like aromatics are also forming SOA (see Sect. 3) although their contribution based on actual understanding of their chemistry has been evaluated to be a small fraction (about 10%) of the global SOA formation in the troposphere (Tsigaridis and Kanakidou, 2003), although locally might be much more important. This contribution is of the same order of magnitude with the naturally driven variation of the SOA chemical production (Tsigaridis et al., 2005).
According to these rough estimates, the chemical forma-tion of SOA may be significant when compared to the pri-mary carbonaceous emissions (about 60% on a global scale and even more regionally).
2.2.4 Uncertainties of estimates
There are large uncertainties associated with both anthro-pogenic and natural emission inventories on regional and global scales. For instance, using detailed land cover and tree species information Guenther et al. (2000) estimated for North America monoterpene emissions of 17.9 and ORVOC emissions of 31.8 Tg y−1. These can be compared with 13.1
and 17.7 Tg y−1presented in Table 2 (for USA and Canada).
A difference of 30 to 80% between these two inventories is deduced. This can not be generalised since other inven-tories might compare better. This is the case for the esti-mate of 6.1 Tg y−1 terpene and OVOC emissions by Simp-son et al. (1999) for OECD Europe that is comparable to the 6.2 Tg y−1given in Table 2. However, when comparing Simpson et al. (1999) in more detail with country specific data compiled by Lenz (2001) differences of a factor of 2 show up over Europe as can be seen from Table 3 where both emission estimates for Italy and for France are compared.
These regional differences and uncertainties propagate to the global scale inventories. As described before, forests have the largest potential to form SOA. Global emissions es-timates of isoprene have an overall uncertainty of a factor of 3 (250-750 TgC y−1, W2004) whereas those of the other
terpenes and sesquiterpenes that are the main known SOA precursors are subject to a factor of 5 uncertainty (W2004). In addition, branch enclosure measurements by Goldstein et al. (2004) confirm more than 100 BVOCs are emitted but not typically observed over the forest; these unmeasured BVOC emissions are approximately 10 times the measured monoterpene fluxes. An extensive overview of the uncertain-ties in these emissions and the global uncertainty range is given by W2004. The main uncertainties are associated with (W2004):
1. tree specific emissions factors and functions
2. the use of geographical data-bases of land-cover, eco-systems and tree abundances
3. foliar density and phenology of these trees 4. environmental conditions.
Most emission rates adopted for the construction of the in-ventories of biogenic emissions and in particular those of monoterpenes and sesquiterpenes have not been evaluated based on measured ambient concentrations. There is a clear need for more data on emissions chemical speciation since aerosol formation potential largely depends on the chemi-cal structure of the precursor molecules. This applies par-ticularly to sesquiterpenes that are known to have the largest potential to form SOA and their emission factors have been poorly studied.
Anthropogenic VOC emissions are 5–10 times lower than biogenic VOC emissions (excluding CH4). When
consid-ering the known SOA precursor emissions alone, this ra-tio increases above 10. Large uncertainties exist also in the anthropogenic emission factors for SOA precursors like aromatics and some oxygenated solvents. Global invento-ries of NMVOC anthropogenic emissions are generally cal-culated in two steps. First, using an emission factor ap-proach, total NMVOC emissions are calculated on the ba-sis of (inter)national activity statistics (e.g. fuel consump-tion, solvent use) and emission factors that take into account
Table 3. Comparison of BVOC emissions – Lenz (2001) and
Simp-son et al. (1999) – for forest tree species in Italy and France (in Gg y−1).
Reference, Area Monoterpenes OVOCs
Lenz (2001), Italy 115 63
Simpson et al. (1999), Italy 32 46
Lenz (2001), France 276 130
Simpson et al. (1999), France 111 110
abatement technologies. According to Olivier et al. (1999b), the uncertainty in total NMVOC emissions has been es-timated to be ∼50% for fossil fuel related emissions and ∼100% for non fossil fuel emissions. Second, the to-tal NMVOC emissions are generally subdivided in specific NMVOC species clusters. Toluene, xylene and trimethyl-benzene are three different NMVOC groups for which an emission profile has been defined in the EDGAR database (Oliver et al., 1996). For each NMVOC profile an activ-ity specific profile has been defined, which assigns the frac-tion of each of the NMVOC group to the total NMVOC emissions. In general global uniform NMVOC profiles are based on data from USA and EU countries (e.g. http:// www.epa.gov/ttn/chief/software/speciate/index.html or http: //reports.eea.eu.int/EMEPCORINAIR3/en). Application of these global aggregated NMVOC profiles leads to another considerable uncertainty. Specific quantitative uncertainty estimates on toluene, xylene and trimethylbenzene are – to our best knowledge – not available at the moment.
2.2.5 Change of Natural Emissions due to land-use and cli-mate change
Global use of land has been changing in the last 2 cen-turies, and is expected to be further modified in the future. The largest recent changes of land-use are in the tropics, mainly due to conversion of tropical forests into crop-lands. For instance from 1970–2000 forest areas in Asia, Latin America and Africa have decreased by 26, 12 and 13%, re-spectively (source: IMAGE2.2; http://arch.rivm.nl/image/). Global forests decreased by 2% in this period. According to the FAO 2003 State of the World Forest Report (http: //www.fao.org/DOCREP/005/Y7581E/Y7581E00.HTM) the decadal deforestation rates for the above mentioned areas were 1, 4, and 8% in the period 1990–2000.
Naik et al. (2004) using a dynamic global ecosystem model calculated that the combined fluctuations in climate and atmospheric CO2 during 1971–1990 caused significant
seasonal (17–25%) and interannual (2–4%) variability in the simulated global isoprenoid fluxes with an increasing trend during this time period.
The quantification of the impact of climate change on fu-ture biogenic VOC emissions remains complex. The most important effect might be an increase in emission rates as a direct result of higher temperatures; however changes in cloudiness, precipitation, and land use may influence emis-sions as well. Little is known about the final combined effect of land-use change and climate change on VOC emissions. An exemplary model study by Sanderson et al. (2002) com-bining climate change and land-use changes suggests that global isoprene emissions may increase by 27%.
2.3 Uncertainties and Research Needs
– The overall knowledge of emissions both of primary
carbonaceous particles and of gaseous precursors of SOA is far from being satisfactory. The overall uncer-tainties range between a factor of 2 and 5.
– Significant effort needs to be put in improving the
BC and OC inventories however progress can be only achieved via standardizing the BC measurements.
– POA appears to be quite an important part of the
car-bonaceous aerosol. However accurate emission esti-mates received relatively little attention. Further stud-ies based on coherent observations are required to con-struct reliable POA emission inventories not only from the anthropogenic sources that are the most commonly considered in modelling studies but also from the bio-genic sources, the importance of which remains to be determined. In particular the potential important POA marine source from the ocean currently omitted from climate modelling prediction should be evaluated (O’Dowd et al., 2004).
– Although sesquiterpenes are known to be the most
ef-ficient natural SOA precursors (see further discussion) their emission rates from vegetation have been poorly studied due to their high reactivity and require more attention in the future with targeted experimental and modelling studies.
– Future studies to improve our knowledge on primary
emissions invoke integration of various approaches to address these questions such as:
– enclosure measurements
– above-canopy and deposition flux measurements of
gases and aerosols
– ambient concentrations measurements
– both forward and inverse modelling to link
emis-sions with observed concentrations
– use of satellite observations in models to
evalu-ate/improve emission inventories or derive them, see for instance examples by Abbot et al. (2003) and Martin et al. (2003).
– The consistency of emission inventories with the
ambi-ent observed concambi-entrations can be evaluated by apply-ing chemistry/transport models to simulate the observed concentrations based on these emission inventories.
3 Representation of Secondary Organic Aerosol
forma-tion in atmospheric models
The processes leading to SOA formation can be viewed as occurring sequentially:
Emissions of gases → Gas-phase chemistry ↔ Nucleation/Gas-particle partitioning ↔ Aerosol-phase/aqueous phase chemistry/cloud processing
To represent SOA formation quantitatively requires each of these steps to be modelled. At present, although the po-tential importance of aerosol-phase chemistry has been re-cently established through the identification of oligomeric species (see Sect. 3.1); these reactions are not yet represented in models.
In this section we first summarize actual knowledge on the chemical reactions responsible for SOA formation (Sect. 3.1) and on the involvement of SOA constituents in nucleation (Sect. 3.2). Then, the aerosol dynamics are summarised (Sect. 3.3) and the gas-to-particle partitioning parameterisa-tions used in atmospheric models (Sect. 3.4) are outlined and finally in Sect. 3.5 the actual representation of SOA forma-tion in current global models is summarized.
3.1 Chemistry of Secondary Organic Aerosol formation
Since monoterpenes would appear to be the major precur-sors of secondary organic particulate matter (SOA) from BVOCs, much work has gone into investigating the reactions of monoterpenes, particularly over the past decade. Sum-maries, up to 2000, of the gas-phase kinetics of the monoter-pene reactions with OH and NO3radicals and ozone,
prod-ucts of these reactions and the pathways leading to their for-mation can be found in several review articles/books (e.g., Atkinson, 1997; Atkinson and Arey, 1998; Calogirou et al., 1999; Calvert et al., 2000). Seinfeld and Pankow (2003) have summarized laboratory studies of SOA formation per-formed over the last decade. The studies have been cate-gorized according to the experimental conditions employed such as NOxphotooxidation, O3reaction or OH reaction.
As a result of ever increasing information on the nature of the gas-phase products and the composition of the resulting aerosol from the oxidation of monoterpenes much effort is now being spent in developing combined gas-phase kinetics and aerosol partitioning models to represent secondary or-ganic aerosol formation in ambient models. (e.g., Kamens et al., 1999; Kamens and Jaoui, 2001; Pankow et al., 2001; Seinfeld et al., 2001; Griffin et al., 2002a, b, 2003; Pun et al., 2002).
Most of the experiments on the atmospheric chemistry of monoterpenes, i.e. phase kinetic rate coefficients, gas-phase product identification and quantification, quantifica-tion of SOA yields and its molecular composiquantifica-tion, have been performed in smog chambers using either natural or artifi-cial sunlight. Seinfeld and Pankow (2003) have discussed the pros and cons of the size of the chamber, indoor and outdoor chambers with artificial and natural light sources, etc and the arguments will not be pursued further here. However, irre-spective of the type of chamber, chamber experiments have inherent difficulties associated with the chemistry of SOA formation when extrapolating the results to atmospheric con-ditions.
– Carefully designed chamber experiments using modern
analytical techniques to allow accurate quantitative de-tection of organics at low concentrations, are needed to investigate SOA formation under atmospheric con-ditions.
In the case of the NOxphotooxidation systems oxidation by
the OH radical will initially dominate, however, as the re-action proceeds, O3and under some circumstances also NO3
radicals will be formed in high enough concentration to com-pete with the OH radical oxidation. This makes assignment of the relative importance of the oxidants OH, NO3and O3in
the SOA formation problematic. In the ozonolysis reactions peroxy radicals will be formed which under most normal at-mospheric conditions would react with NO to form alkoxy radicals. Ozonolysis experiments can not be performed in the laboratory in the presence of NOxsince the NOxreacts
rapidly with O3.
– It is, therefore still an open question as to whether the
aerosol yields observed in laboratory ozonolysis exper-iments are transferable to atmospheric conditions; in re-ality the yields could be higher or lower.
Recently, Docherty and Ziemann (2003), Ziemann (2003) and Keywood et al. (2004) have shown that the presence of OH scavenger in ozonolysis experiments has significant im-pact on SOA yields. Keywood et al. (2004) explained this be-haviour by the involvement of acylperoxyradicals formed via isomerisation of alkoxy radical which in turn are produced from Crieege Intermediates during ozonolysis of endocyclic alkenes. Winterhalter et al. (2000), Koch et al. (2000) and Jenkin et al. (2000) showed that acylperoxy radicals from both endo- and exocyclic monoterpenes are involved in the formation of dicarboxylic acids via permutation reactions with HO2or RO2radicals. These radical reactions are
ini-tiated by the decomposition of the excited Criegee Inter-mediates via the hydroperoxy channel or the ester channel (Calvert et al., 2000). Bonn et al. (2002) have recently shown that the addition of H2O and carbonyl compounds affects the
yield of SOA. This observation was explained by the involve-ment of stabilized Criegee Intermediates in the SOA forma-tion processes.
The translation of these results to the real atmosphere re-quires thorough interpretation and further understanding of the corresponding chemical mechanisms. For the reaction of NO3 with monoterpenes this may also be an issue but
be-cause the reactions occur mainly during the night time when NO is low the effect (if any) will not be so far reaching.
– In NO3 radical chamber chemistry systems, however,
reactions of the high levels of NOxoften employed can
block many reaction pathways, which would otherwise be important under atmospheric conditions.
3.1.1 Gas phase reactions leading to semivolatile products
Because of the potential importance of monoterpenes to SOA formation much of the research related to elucidating the ox-idation mechanisms and products of monoterpenes known to be emitted into the troposphere in substantial quantities (Atkinson and Arey, 1998; Calogirou et al., 1999) has been focussed primarily on α- and β-pinene. Measurements of monoterpenes speciation suggest that these make a particu-larly significant contribution to global monoterpenes emis-sions (Guenther et al., 1994; Geron et al., 2000 and refer-ences therein). These monoterpenes are also representative of classes of monoterpenes having either an endocyclic dou-ble bond (in the case of α-pinene) or an exocyclic doudou-ble bond (in the case of β-pinene), therefore, the discussion on monoterpenes chemistry leading to SOA will focus on these two compounds and, in particular, on α-pinene.
In the case of α-pinene considerable progress has been made in determining the kinetics and elucidating the mecha-nisms of the early stages of its gas-phase degradation chem-istry initiated by reaction with OH radicals (e.g., Arey et al., 1990; Hakola et al., 1994; Hallquist et al., 1997; Vinckier et al., 1997; Aschmann et al., 1998; Noziere et al., 1999a; Orlando et al., 2000; Jaoui and Kamens, 2001; Larsen et al., 2001; Wisthaler et al., 2001; Winterhalter et al., 2003), NO3
radicals (e.g., Wangberg et al., 1997; Berndt and B¨oge, 1997; Hallquist et al., 1997; Jang and Kamens, 1999) and ozone (e.g., Hakola et al., 1994; Alvarado et al., 1998a; Kamens et al., 1999; Yu et al., 1999a; Koch et al., 2000; Orlando et al., 2000; Winterhalter et al., 2003). The further oxidation of the major first generation product, pinonaldehyde, has also been studied quite extensively (e.g., Glasius et al., 1997; Hallquist et al., 1997; Alvarado et al., 1998b; Noziere et al., 1999a, b; Jaoui and Kamens, 2003a). There have also been some the-oretical studies of the oxidation mechanisms (Peeters et al., 2001; Vereecken and Peeters, 2000).
The reactions with OH, NO3 and O3 lead to a large
suite of oxygenated reaction products which include alde-hydes, oxy-aldealde-hydes, carboxylic acids, oxy-carboxylic acids, hydroxy-carboxylic acids, dicarboxylic acids, or-ganic nitrates etc. In addition, several peaks with m/z 187 [M+H]+ observed in the LC-MS analysis of fil-ter samples from the oxidation of α-pinene (Hoffmann,
Table 4. Structures and IUPAC names of a selection of products typically observed in the oxidation of α-pinene (source: Winterhalter et al., 2003). 1 2 3 4 5 OH O O OH O CHO O O OH COOH O 2-hydroxy-3-pinanone 8-hydroxy-menthen-6-one
pinonaldehyde pinalic acid pinalic acid
2-Hydroxy-2,6,6-trimethyl-bicyclo[3.1.1]heptan-3-one ethyl)-2-methyl-cyclohex-2- 5-(1-Hydroxy-1-methyl-enone
(3-Acetyl-2,2-dimethyl-cyclobutyl)-acetaldehyde ethyl)-cyclobutane-carboxylic 2,2-Dimethyl-3-(2-oxo-acid (3-Formyl-2,2-dimethyl-cyclobutyl)-acetic acid 6 7 8 9 10 O COOH COOH OH COOH COOH O CHO O O O O norpinonic acid Mw 172 "pinolic acid" norpinic acid 10-keto-pinonaldehyde 4-keto-pinonaldehyde 3-Acetyl-2,2-dimethyl-cyclobutanecarboxylic acid 3-(2-Hydroxy-ethyl)-2,2- dimethyl-cyclobutane-carboxylic acid 2,2-Dimethyl-cyclobutane-1,3-dicarboxylic acid (3-Acetyl-3-oxo-2,2- dimethyl-cyclobutyl)-acetaldehyde (3-Acetyl-2,2-dimethyl- cyclobutyl)-2-oxo-acetaldehyde 11 12 13 14 15 O OH CHO O CHO HO O CHO OH O O CHO O O O CHO
10-OH-pinonaldehyde 1-OH-pinonaldehyde 4-OH-pinonaldehyde pinalic acid methyl ester [3-(2-Hydroxy-ethanoyl)-2,2- dimethyl-cyclobutyl]-acetaldehyde (3-Acetyl-3-hydroxy-2,2- dimethyl-cyclobutyl)-acetaldehyde (3-Acetyl-2,2-dimethyl- cyclobutyl)-2-hydroxy-acetaldehyde 3-Acetyl-5,6-dioxo-heptanal 2,2-Dimethyl-3-(2-oxo-ethyl)-cyclobutanecarboxylic
acid methyl ester
16 17 18 19 20 O O CHO O O O H O COOH OOH OH OH OOH
pinonic acid ββββ
-hydroxy-hydroperoxide ββββ -hydroxy-hydroperoxide Acetic acid
2,2-dimethyl-3-(2-oxo-ethyl)-cyclobutyl ester dimethyl-cyclobutyl ester Formic acid 3-acetyl-2,2- (3-Acetyl-2,2-dimethyl-cyclobutyl)-acetic acid 2-Hydroperoxy-2,6,6- trimethyl-bicyclo[3.1.1]heptan-3-ol 3-Hydroperoxy-2,6,6- trimethyl-bicyclo[3.1.1]heptan-2-ol 21 22 23 24 25 COOH COOH C(O)OOH CHO O COOH O O COOH O O O CHO OH
pinic acid pinalic-peroxo acid 7-keto-pinonic acid 4-keto-pinonic acid 3-Carboxymethyl-2,2- dimethyl-cyclobutane-carboxylic acid 2,2-Dimethyl-3-(2-oxo- ethyl)-cyclobutane-carboperoxoic acid [2,2-Dimethyl-3-(2-oxo-ethanoyl)-cyclobutyl]-acetic acid
(3-Acetyl-2,2-dimethyl-cyclobutyl)-2-oxo-acetic acid ethyl)-5,6-dioxo-heptanal
3-(1-Hydroxy-1-methyl-26 27 28 29 30 O COOH OH O COOH HO O COOH OH O CHO OOH C(O)OOH COOH
10-OH-pinonic acid 1-OH-pinonic acid 4-OH-pinonic acid
4-hydroperoxy-pinonaldehyde peroxo-pinic acid [3-(2-Hydroxy-ethanoyl)-2,2-dimethyl-cyclobutyl]-acetic acid (3-Acetyl-3-hydroxy-2,2-dimethyl-cyclobutyl)-acetic acid (3-Acetyl-2,2-dimethyl-cyclobutyl)-2-hydroxy-acetic acid (3-Acetyl-2,2-dimethyl- cyclobutyl)-2-hydroper-oxy-acetaldehyde 3-Hydroperoxycarbonyl- methyl-2,2-dimethyl-cyc-lobutane-carboxylic acid
2001; Winterhalter et al., 2003) have been tentatively as-signed to hydroperoxides; 2-hydroperoxy-3-hydroxypinane (2-hydroperoxy-2,6,6-trimethyl-bicyclo [3.1.1] heptan-3-ol) and 3-hydroperoxy-2-hydroxypinane (3-hydroperoxy-2,6,6-trimethyl-bicyclo [3.1.1] heptan-2-ol). Two isomers ex-ist of each compound, so in principle four products may
be present. Peroxo-pinalic acid (2,2-dimethyl-3-(2-oxo-oxoethyl)-cyclobutane-carboperoxoic acid) is also a possible product candidate. Table 4 gives the structures and IUPAC names of 30 of the products observed in the oxidation of a-pinene.
– There is mounting evidence from laboratory studies of
monoterpenes oxidation by O3, OH- and NO3-radicals
that the most important process with regard to aerosol formation is the reaction with ozone (Hoffmann et al., 1997; OSOA project: Hoffmann, 2001).
– There is very little and fragmentary information
avail-able about the SOA yields from sesquiterpenes. These yields are much higher than those of the monoterpenes (17–67% on a mass basis reported by Griffin et al., 1999b) and based on these yields the contribution of sesquiterpenes to SOA global formation may be up to 9% (Griffin et al., 1999a).
Oxidation products of pinenes others than those detected dur-ing ozonolysis chamber experiments have been observed in the ambient SOA suggesting that other oxidants and sec-ondary reactions may be involved in oxidized SOA forma-tion (Claeys et al., 2004b; Edney et al., 2003; Kub´atov´a et al., 2002).
The contributions of the three major oxidation processes of monoterpenes (OH, NO3and ozone) to new aerosol
forma-tion and aerosol yield have been found in laboratory studies to be very different for the three possible reactions.
– Ozone was found by far to have the highest potential to
form new particles at similar reactant consumption rates of α-pinene as well as for β-pinene (Bonn and Moort-gat, 2002).
– Reactions of the monoterpenes with either OH or NO3
result in remarkably less nucleation but with minor dif-ferences in the new-formed aerosol volume compared to the ozone reaction. This suggests that products with higher volatility than in the ozone experiments may have been formed in OH and NO3reactions (Hoffmann,
2001).
For new particle formation to occur by homomolecular nu-cleation, an oxygenated product must be generated in the gas phase at a concentration in substantial excess of its sat-uration vapour concentration with respect to the condensed phase. Oxygenated products capable of generating new parti-cles in the atmosphere must of necessity be particularly non-volatile. Inclusion of polar functional groups with retention of carbon number will reduce product volatility quite con-siderably in comparison with that of the parent hydrocarbon. Experimental evidence indicates that high molecular weight compounds containing the –OH, -C=O and, in particular, the carboxylic acid -C(=O)OH functionality are particularly im-portant in this respect (e.g., Tao and McMurry, 1989; Yu et al., 1998 and references therein). The ability of most SOA compounds to nucleate in the atmosphere is not well under-stood and is the topic of current research. One needs to un-derstand not only the volatility of these compounds but also their surface energy because of the importance of the Kelvin
effect for nucleation. Note, that the increase in molecular weight leads to greater Kelvin effect (Seinfeld and Pandis, 1998) that hinders the homogeneous nucleation. Therefore, for an organic species to homogeneously nucleate, the low-ering of vapour pressure when increasing functionality with increased molecular weight should overcome the counteract-ing Kelvin effect.
Until recently, dicarboxylic acids were the lowest volatile compounds positively identified in terpenes generated aerosol. cis-Pinic acid (a C9 dicarboxylic acid) has been
identified as a condensed product of the ozonolysis of both α- and β-pinene (e.g., Christoffersen et al., 1998; Hoffmann et al., 1998; Glasius et al., 2000; Kamens et al., 1999; Jaoui and Kamens, 2003b, c).
It has been suggested (Koch et al., 2000; Winterhalter et al., 2000; Jenkin et al., 2000) that cis-pinic acid is the most likely photooxidation product of both α- and β-pinene that will result in prompt formation of new aerosols by nucle-ation. Pathways leading to cis-pinic acid have been sug-gested from the secondary reactions of the first-generation gas-phase products generated in monoterpenes photooxida-tion (Jenkin et al., 1997) or from their auto-oxidaphotooxida-tion in the condensed phase (Jang and Kamens, 1999). However, as remarked by Jenkin et al. (2000) the observed timescale of aerosol formation (Koch et al., 1999) appears to require that cis-pinic acid is itself a “1st-generation product”. Winterhal-ter et al. (2000), Koch et al. (2000) and Jenkin et al. (2000) have suggested a possible mechanism for the formation of cis-pinic acid from the ozonolysis of both α- and β-pinene. These mechanisms have been discussed by Jenkin (2004). The key intermediate is an acyl radical (see circled interme-diate in Fig. 2, adopted from Winterhalter et al., 2000), which is formed from exo- and endocyclic alkenes, like α- and β-pinene. Two pathways are proposed for this acyl radical. Ei-ther isomerisation of the complex C9-acyloxy radical by an
1,7 H atom shift (see Fig. 2, Pathway A: Jenkin et al., 2000) or reaction with HO2, yielding pinalic peroxo acid, which
finally isomerizes to cis-pinic acid (Fig. 2: Pathway B: Win-terhalter et al., 2000).
– There is, however, evidence that compounds less
volatile than dicarboxylic acids are present in the aerosol.
Edney et al. (2003) and Kub´atov´a et al. (2002) have detected a C8 tricarboxylic acid, an α-pinene oxidation product, in
semi-rural and urban aerosols. Ziemann (2002) has presented evidence for the possible formation of diacyl peroxides in the ozonolysis of cyclohexene and homologous compounds and has suggested that these compounds may be the major nucleating agent in these systems and are also responsible for a significant fraction of the aerosol mass.
Mechanisms have been developed to describe the reaction pathways leading to these products. For example, explicit or lumped degradation schemes can be found for the chem-istry of both α- and β-pinene either in the new version of the
2 2 2 2 2 2 2 2 2 2 2+ 2 2 2 2 2 2+ 2 2 2 2 2 2 2 0 +2 0 +2 2+ 2 2 2+ 22 +2 SLQRQDOGHK\GH 2 &+2 2 2 2 2 2+ 2 2 2+ 2+ &22+ 2 SLQRQLFDFLG +2 2 2 2 + 2 2 SLQRQDOGHK\GH + 2 DSLQHQH SULPDU\R]RQLGH 2 2 22 2 2 22+ 2 2 2+ 2 2 2 2 2 2 2 2 +&2 QRUSLQRQDOGHK\GH 2+ +2 2 52 12 2 2[ &22+ 2 QRUSLQRQLFDFLG NHWRSLQRQDOGHK\GH 2+SLQRQDOGHK\GH 2 2 22 2 2 22 2 2 +2 2 2 2 12 52 2+SLQRQDOGHK\GH 2 2 &+&2 2 2 2 2+ 2 2+ 2 2 2 22+ 2 2 2 2 2 2+ 12 52 2+SLQRQDOGHK\GH 2 2 +&+2 2 2 2 NHWRSLQRQDOGHK\GH 2 2 2 22 2 2 2 2 2 22+ &22+ &22+ SLQLFDFLG SLQDOLFSHUR[R DFLG 2 2 2+ SLQDOLFDFLG 2 2 2+ 2 2+ 22 2 2 12 +2 2 +2 2 +2 2 +2 2 &22+ &22+ SLQLFDFLG 2 2+ SHUR[RSLQLFDFLG 2 22+ +\GURSHUR[\GHFKDQQHO6 &ULHJHH,QWHUPHGLDWHV (VWHUFKDQQHO (VWHUFKDQQHO +3 +3 +3 DF\OUDGLFDO +2
$
%
Fig. 2. Reaction mechanism for the ozonolysis of α-pinene (adopted from Winterhalter et al., 2003). The two Criegee Intermediates and
the main decomposition channel (Hydroperoxyde channel) are highlighted with square boxes. One channel leads to the formation of an acyl-type radical (circle), which can also be formed in case of β-pinene. Consecutive reactions of this acyl radical then lead to cis-pinic acid (also highlighted by circles) via two possible pathways (A: Jenkin et al., 2000, B: Winterhalter et al., 2000).
Master Chemical Mechanism (MCM version 3: Saunders et al., 2003; http://www.chem.leeds.ac.uk/Atmospheric/MCM/ mcmproj.html) or SAPRC-97 and SAPRC-99 (Carter, 1997, 1999). Figure 2 shows a schematic overview of the possi-ble pathways in the ozonolysis of α-pinene leading to prod-ucts which have been observed experimentally (Winterhalter et al., 2003) indicating also the formation of hydroxyperox-ides. Recent effort by Jenkin (2004) to model the formation and composition of SOA produced during the ozonolysis of pinene using the MCM mechanism (v3) is also pointing to
the key role of multifunctional products of VOC oxidation in SOA formation which contain, for example, the hydroperox-ide functionality. Bonn et al. (2004) have also pointed out the importance of hydroperoxides in global SOA formation.
3.1.2 Organic polymerisation in the aerosol phase
A long-standing puzzle associated with the analysis of molecular speciation of SOA has been the presence in the aerosol of species whose vapour pressures are far too high to support significant partitioning into the aerosol phase
Fig. 3. Dimerisation of pinonaldehyde via aldol condensation (upper panel) and by gem-diol formation with subsequent dehydration (lower
panel) (Tolocka et al., 2004).
(Forstner et al., 1997ab; Yu et al., 1998, 1999a, b). It was speculated that these relatively small and volatile species might actually be decomposition products of larger, less volatile molecules that were broken apart by the relatively harsh environment of the mass spectrometric methods tra-ditionally used for analysis. Very recently, high molecular weight (and therefore low vapour pressure) products have been identified in the aerosol phase using analytical tech-niques that do not tend to break the molecules apart (Kalberer et al., 2004; Tolocka et al., 2004; Gao et al., 2004). Ex-perimental results are just now emerging, but the existence of heterogeneous reactions between semi-volatile condensed SOA products to yield compounds of much lower volatility could play an important role in causing SOA yields to ex-ceed those calculated solely on gas-particle partitioning of the gaseous semi-volatile oxidation products.
Oligomer and/or polymer formation following both bio-genic and anthropobio-genic VOC degradation has been proven and may be considered as responsible for an important fraction of the SOA chemical build up in the troposphere (Tolocka et al., 2004; Kalberer et al., 2004; Gao et al., 2004). Limbeck et al. (2003) have shown that SOA formation of atmospheric polymers – humic like substances – occurs by heterogeneous reaction of isoprenoid and terpenoid emission in the presence of a sulphuric acid aerosol catalyst. Jang et al. (2002, 2003) presented chemical mechanisms leading to the formation of low volatility organic products from oxi-dation of aldehydes by acid-catalysed heterogeneous reac-tions. Iinuma et al. (2004) and Gao et al. (2004) found that acid catalysis can increase by about 40% the particle phase organics produced during ozonolysis experiments and suggest that condensation of smaller molecules takes place by polymerization or aldol condensation following the
for-mation of aldehydes from terpenes ozonolysis. Tolocka et al. (2004) and Gao et al. (2004) have identified oligomers as large as tetramers in chamber experiments during α-pinene ozonolysis using acidic inorganic seed aerosol. They con-clude that α-pinene ozonolysis in the presence of an acid catalyst is strongly influenced by oligomerisation reactions of primary ozonolysis products, most likely by aldol conden-sation and/or gem-diol formation (Fig. 3). An ion consistent with the dimer of pinonaldehyde has been also detected on ambient aerosols (Tolocka et al., 2004). These recent dis-coveries provide a new point of view for SOA formation ex-perimental studies and modelling. The exact mechanisms of the oligomer and/or polymer formation and their significance for the chemical formation and properties of the secondary organic aerosol remain to be determined.
3.1.3 Multiphase oxidation of hydrated gases
Claeys et al. (2004a) have analysed aerosols from the Ama-zonian rain forest and identified in the fine size fraction con-siderable quantities of a mixture of two diastereoisomeric 2-methyltetrols, 2-methylthreitol and 2-methylerythritol, which they proposed could be explained by isoprene oxida-tion. Note that, until recently, isoprene was not considered a significant contributor to SOA.
The first paper was followed very quickly by a second, Claeys et al. (2004b), in which they revise the mechanism postulated in Claeys et al. (2004a) by which 2-methyltetrols are generated from isoprene. In their latest study they pro-pose the multiphase acid-catalysed oxidation of isoprene, methacrolein and methacrylic acid with hydrogen peroxide as a new route to SOA formation. They mention that par-titioning of isoprene into the aqueous phase is expected to be enhanced under acidic conditions since isoprene is hy-drated in aqueous solutions of sulphuric acid (Ryabova et al., 1992). Based on preliminary kinetic studies, Claeys et al. (2004b) suggest that these SOA forming reactions are more likely to occur in hydrated aerosols or haze droplets rather than in short-lived cloud droplets. Finally, they sug-gest that monoterpenes and their gas phase oxidation prod-ucts might undergo similar multiphase reactions leading to SOA formation.
This proposed multiphase mechanism, reaction with hy-drogen peroxide under acidic conditions is an analogue to atmospheric sulphate formation and is different from all pre-viously mentioned mechanisms. The importance of this new route in the global SOA formation requires investigation; however, this will only be possible when an improved under-standing of the actual mechanism and the associated kinetic data become available.
3.1.4 Concluding remarks
Over the past decade significant progress has been made in our understanding of the gas-phase oxidation mechanisms of
biogenic hydrocarbons. However, our knowledge is far from being complete and many gaps exist in the determination
– of all potential gaseous anthropogenic and biogenic
pre-cursor molecules of SOA,
– of the secondary photooxidation processes in the gas
phase leading to low volatility compounds and thus to SOA formation,
– of the impact of NOxlevels on the final products of the
SOA formation chemistry
– of the heterogeneous reactions between particle
associ-ated substances and gaseous compounds able to modify the composition and the mass of aerosol,
– of the aerosol chemistry responsible for the recently
re-ported oligomer formation which increases the aerosol mass,
– of the aqueous phase chemistry, which might produce
semi-volatile compounds that build up aerosol mass,
– of the complete molecular composition of the aerosol
produced from the above mentioned processes.
A full appreciation of all the processes involved will proba-bly have to await the evolution of new more powerful ana-lytical probing techniques that are starting to emerge. Fur-ther experimental work is also needed in order to translate laboratory results obtained in high concentrations of organic precursors and oxidants to atmospheric conditions with often much lower concentrations.
3.2 Nucleation
The ability of SOA compounds to form new particles in the atmosphere is a rather controversial issue. In the laboratory, formation of new particles is routinely observed during oxi-dation of SOA precursors (Stern et al., 1987; Hatakeyama et al., 1989; Pandis et al., 1991; Wang et al., 1992; Hoffmann et al., 1998; Koch et al., 2000; Hoppel et al., 2001; Bonn et al., 2002). However, these experiments often use moderate to high concentrations of VOCs and no pre-existing aerosol. For example, reaction of 20 ppb of α-pinene with 120 ppb of ozone and no pre-existing particles can create 50 000 cm−3
(Hoppel et al., 2001). Nucleation in these systems (even at high concentrations) can be relatively easily suppressed if a moderate concentration of seed aerosol is present (Cocker et al., 2001a, b). In this case the SOA compounds condense on the pre-existing aerosols and nucleation does not take place at the same time. Many studies have reported rather frequent nucleation events in the boundary layer and free troposphere in a variety of environments (see Kulmala et al., 2004a for a review of field observations). It has been suggested that
some of these observed new particle formation events in re-mote areas may be due to nucleation of biogenic SOA com-pounds (Marti et al., 1997; O’Dowd et al., 2002; Kavouras and Stephanou, 2002).
The potential for the formation of new particles during var-ious reactions of α- and β-pinene was investigated by Bonn and Moortgat (2002). Their experiments indicated that the ozonolysis dominates the new particle formation compared to the reactions with OH and NO3. The authors argued that
ozonolysis is probably the only atmospherically relevant or-ganic source for new particle formation from biogenic pre-cursors, because of the low atmospheric concentrations of these biogenic compounds. The formation of new particles during the monoterpenes ozonolysis is negatively affected by water vapour (Bonn et al., 2002). Bonn and Moort-gat (2002) suggested the involvement of stabilized Criegee Intermediates and the formation of secondary ozonides as nu-cleating species. Bonn and Moortgat (2003) argued that the atmospheric new particle formation observed in remote ar-eas and generally attributed to low-volatility oxidation prod-ucts of monoterpenes may be actually initiated by prodprod-ucts of sesquiterpenes reactions with ozone. The cyclic alkene ozonolysis experiments of Ziemann (2002) suggested that di-acyl peroxides and not dicarboxylic acids might be the nucle-ating agents in SOA systems. Hoppel et al. (2001) concluded that the nucleating species during α-pinene ozonolysis must have a saturation mixing ratio lower than 10 ppt at 25oC and that classical nucleation theory is not able to explain their observations. For the anthropogenic emissions the processes leading to nucleation and SOA will probably be much more varied than for the biogenics because of the greater diversity in the chemical structures of the compounds emitted. There are experiments (Johnson et al., 2004b) which support that ozone is not a major player in the aromatic systems and that nitroaromatic compounds may be playing a dominate role. For anthropogenic species containing double bonds, how-ever, the evidence here again (Tobias et al., 2000; Kalberer et al., 2000) supports that ozone reactions are probably the most important for SOA formation.
The analysis of all the measurements in a boreal forest site in Southern Finland indicated that the most probable new particle formation mechanism in the area is ternary nucle-ation of water-sulphuric acid-ammonia (Kulmala et al., 2001; Janson et al., 2001). After nucleation the major part of the growth is probably due to condensation of organic vapours. However, there is lack of direct proof of this phenomenon be-cause the composition of 1–5 nm size particles is extremely difficult to determine using the present state-of-art instru-mentation. Similar conclusions have been reached by Marti et al. (1997) about new particle formation in Colorado and by Gaydos et al. (2005) about the nucleation events in the NE US. Gao et al. (2001) also concluded based on their smog chamber studies that sulphuric acid is a superb nucleating species, while secondary organic compounds probably play a role more confined to growing newly formed particles. The
dominance of sulphuric acid as a nucleating agent over atmo-spheric organics was also suggested by the work of Tobias et al. (2001). The authors investigated the formation of new particles in the exhaust of a diesel engine and concluded that the results were consistent with a mechanism of nanoparti-cle formation involving nunanoparti-cleation of sulphuric acid and wa-ter, followed by particle growth by condensation of organic species.
Kulmala et al. (2004a) in their review of the available field observations of nucleation concluded that organic vapours could, in principle, participate in nucleation, but nucleation mechanisms that involve organics have not yet been identi-fied. It appears very likely, however, that organics contribute to the growth of nucleated particles and indirectly affect the formation rate of new particles of detectable sizes (Zhang and Wexler, 2002; Anttila and Kerminen, 2003; Kulmala et al., 2004b). Boy et al. (2003) estimated that condensation of monoterpenes oxidation products is able to explain 10–50% of the observed growth rates of fresh particles in Southern Finland.
Garman et al. (2004) investigated theoretically the bi-nary homogeneous nucleation of water-succinic acid and water-glutaric acid based on the classical nucleation theory. They concluded that under atmospheric conditions these bi-nary systems would not form new particles. Kavouras and Stephanou (2002) measured, by using a minimizing artifact sampling device, biogenic primary organic polar compounds and monoterpene carbonyl and acidic photooxidation prod-ucts in both gas and particles over a Mediterranean conifer forest. On the basis of these field measurements they cal-culated saturation concentrations of the acidic and carbonyl photooxidation products for non-ideal conditions using a pre-viously developed absorptive model (Pankow, 1994). The results of this study suggested that the formation of SOA goes on through a heterogeneous heteromolecular nucleation mechanism, where the effects of both pre-existing organic aerosol (mostly primary) and ambient temperature are cru-cial.
The absence of nucleation events in the Amazon (Andreae, Swietlicki personal communication), an environment with very low sulphur concentrations but very high biogenic VOC concentrations, strongly suggests that nucleation of purely SOA compounds in the atmosphere may not be that fre-quent. A potential explanation for this could be that most low-vapour-pressure organics have high molecular weight, but high-MW species have a strong Kelvin effect at nucle-ation sizes as discussed in Sect. 3.1.1.
Zhang et al. (2004) based on laboratory experiments of mixtures of aromatic acid vapours (benzoic and p- and m-toluic acids) with sulphuric acid (H2SO4)have shown that
the presence of these organic acids enhance sulphuric acid nucleation. Based on bonding energies theoretical calcula-tions, they explain these finding by the formation of aromatic acid- sulphuric acid complexes via two hydrogen bonds. The organic acid molecule acts as both a hydrogen bond donor
![Table 2. Regional breakdown of Anthropogenic Primary Organic Aerosol, Black Carbon, terpenes (excluding isoprene), other reac-tive VOC (ORVOC) terpenes [Tg y −1 ]](https://thumb-eu.123doks.com/thumbv2/123doknet/13239873.395304/7.892.78.422.180.461/regional-breakdown-anthropogenic-primary-organic-aerosol-terpenes-excluding.webp)
